A descriptive cross sectional study, using the logical framework approach to conduct a process evaluation was conducted. This study was conducted in two contrasting districts of Mashonaland East province. Seke had the best performance results whereas Goromonzi District had the least, despite comparable geography and support, hence the comparative evaluation. Quantitative data and qualitative key informant interview data were collected concurrently by the investigator after pretesting and relevant readjustments.

The study recruited all 44 trained and active implementers of the VMMC program in Seke and Goromonzi districts. The entire population of nurses and doctors at health facilities offering VMMC were recruited into the study. The District Medical Director (DMO) and the District Nursing officer (DNO) from the district hospitals were purposively recruited as key informants. A total of seven health facilities offering VMMC services in the two districts were visited in Seke and Goromonzi districts.

Permission to conduct the study was obtained from the Mashonaland East Pro- vince Institutional review Board. Written informed consent was obtained from all study participants. Confidentiality was maintained by avoiding the use of individual identifying information and keeping all data under a password controlled computer.

Data were collected concurrently between 9^th and 20^th of January 2017 by the investigators. An interviewer-administered questionnaire and checklists were used to assess for inputs, processes, outputs and progress towards outcome evaluation. A desk review of VMMC records was also done. The questionnaires were coded and entered into the computer. Data cleaning was performed to check for errors in data entry prior to analysis.

Data were analysed using Epi Info ^TMStatistical Package version 7. Frequencies and proportions were generated. In-depth data collection was achieved through face to face interviews. Qualitative data was analysed manually for content.

The total number of participants, purposively recruited was 44, with an equal number coming from the two districts. In total, the majority (65.9%) were nurses. Zimbabwe national army (ZNA) nurses constituted 9% of the participants and were from Seke district alone. The median years of service were 5 years (Q₁ = 3; Q₃ = 25) for Seke and 9 years (Q₁ = 6; Q₃ = 11) for Goromonzi district. The District Medical officers (DMO’s) and their District Nursing officers (DNO’s) for the two districts were purposively recruited as key informants (Table 1).

The number of health workers for the VMMC program was 93% for Goromonzi and 80% for Seke district, out of the required 15 per district. Both districts had two teams each for VMMC implementation. Goromonzi had two more nurses

(14) as compared to Seke which had 12. Both Seke and Goromonzi had one vehicle each for service delivery and none for demand creation. However, Seke district executive provided a second vehicle, from their own district pool, to assist the program. In 2016, both districts received adequate resources for successful implementation of the program with no recorded stock outs throughout the period. No promotional materials were received by the 2 districts in 2016 (Table 2).

Training workshops that were conducted in and outside the province resulted in a balanced number of trained personnel (24) being distributed between the 2 districts. Both districts reported conducting review and planning meetings (4) which were more that the targeted number (3). Notably, both district held 1 out of the provisioned 4 adverse events review meetings, despite Goromonzi having experienced a high number of moderate to severe adverse events following circumcision (55). Inadequate (6/12) mobilisations were recorded in Goromonzi. Notably, Seke received support (3 additional nurses) for outreach. The support of ZNA nurses came from Manyame Airbase (Table 3).

Both districts conducted one more campaign than had been targeted. Seke district circumcised 99.5% (4716) of their set target whereas Goromonzi managed to circumcise 48.5% (2372) of their set targets. The targets are varied, according to previous performance and the population base of the district. Moderate to severe adverse events following circumcision recorded were 0.04% (2) for Seke district and 2.3% (55) for Goromonzi district. Adverse events are required to be

maintained at below 2% according to the service guidelines. Client follow-ups were done during reviews by both districts in accordance with the set targets (Table 4).

Out of their district targets to contribute to 1.3 million circumcisions of HIV-negative males nationally by 2018, from January 2009 to December 2016, Seke district circumcised 83% of HIV negative, target population males (10 - 29-year-olds), realising a discrepancy performance of only 17%. On the other hand, Goromonzi district achieved 15% circumcisions for the same population and period, realising a discrepancy of 85% circumcisions to be done by 2018.

Twenty-one (95.5%) participants in Seke district stated that program ownership by the district management influenced above average performance of the program. Effective coordination mechanisms and implementing motivation by the health workers implementing the program was cited by 22 (100%) and 20 (90.9%) of the participants, respectively. Fifteen (68.2%) of the participants

thought effective mobilisations, whereby Environmental health technicians and ZNA nurses conducted community mobilisation could have enhanced program performance in Seke.

The majority of the participants (90.9%) in Goromonzi stated that late payments by the sponsoring partner were the major reason for the poor performance of the program. Lack of effective coordination of the VMMC activities within the district was identified as the second (77.3%) most prevalent reason for the underperformance of the program. Weak mobilisation strategies and inadequate transport were the least common reasons for program performance at 50% and 36.4% respectively. The fact that VMMC is not integrated into the health facility departments was cited by 63.6% of the participants as one of the reasons for poor performance.

Resource supply for execution of VMMC was adequate, with no reported stock outs. However, this did not translate to standard performance for Goromonzi, contrary to the findings by Gray et al. 2004 in Uganda where a direct relationship between adequate resourcing and optimal performance was observed [4] . The two districts were comparable regarding material resource supply for the period reviewed, the number of teams implementing the program, geography and administration. The two districts were financially supported from inception by the same partner (PSI) and had disbursements done simultaneously.

The fact that Seke received assistance from ZNA nurses for their onsite and outreach activities put them at an advantage over Goromonzi. Seke district received additional 3 nurses from Manyame air base and this provided human resources, solely for the purpose of conducting circumcisions. Health workers implementing the program are full employees at their respective institutions and need to balance routine responsibilities and VMMC activities. Though Goromonzi enjoyed an increased number of trained doctors and nurses over Seke district, however, this marginal advantage did not translate to better performance of the program. Adequate human resource supply was identified as a key aspect to program success, according to Fleming et al. (2008) in Brazil [13] .

The absence of promotional materials in both districts poised a disadvantage in both districts. This was consistent with findings by Weiss et al. (2011) which indicated the key role of promotional material in demand creation in a meta- analysis for 4 southern African countries [12] .

Out of a target of 20, Goromonzi had only 6 active community-based mobilisers who were not evenly distributed within the catchment community. Other villages are reported to have more than 1 cadre whilst others do not have. This resulted in weak mobilisation activities. Community-based mobilisers are an effective strategy that utilises peers within the same community to mobilise them for the program and this is working effectively in Seke.

High incidence rate of moderate to severe adverse events following circumcision in Goromonzi suggests that not only is the district failing to circumcise clients as expected, but that the few being circumcised are not being done so in accordance with the set standard operating procedure. Goromonzi conducted one adverse event review meeting despite unacceptably high incidences. Above 2% adverse moderate to severe adverse events following circumcision is a serious cause for concern, particularly with the backdrop of poor performance.

Though the 2 districts received one vehicle each for both service delivery and demand creation activities, Seke district executive provided the second vehicle in support of VMMC activities which was not the case with Goromonzi. This revealed program ownership on the part of Seke. Goromonzi cited inadequate transport availability as one of the hindrances to program success in. The success of any program is largely depended on the buy-in of the community and its leadership. Some of the interviewed members of the Goromonzi District health executive (DHE) were not aware of the VMMC activities and available resources such as the partner provided the vehicle. This is also supported by 63.6% of the participants who cited lack of program ownership as one of the reasons for poor performance.

Inadequate coordination, as evidenced by minimal consultation between the district VMMC coordinator and the DHE in Goromonzi and suboptimal sharing of the available vehicle between the two teams worked against the program. Members of the DHE interviewed revealed that they received no feedback from the coordinator when the team went out for outreach or campaigns. They, therefore, remained unaware of the success and failures of the program. The ideal situation is for the district coordinator, appointed by the DHE, to always update on VMMC plans as well as yields following outreach and campaigns.

A generally demotivated attitude by the health workers, cited by 54.5% of the participants was apparent too. The chronic underperformance did not seem to trigger concern as evidenced by the 2 teams in Goromonzi not doing much to explore solutions. Given the poor results, it was expected that the district should be holding crisis meetings in order to identify drawbacks and explore ways to enhance performance.

VMMC is implemented by health workers who have other activities that they need to prioritise following allocation on a daily basis. VMMC is not recognised as part of the service centre departments that is given personnel. This seemed to work against the program in both districts since workers would need relief from routine duties to attend to VMMC clients. In a study by Ashengo et al. in Tanzania, integration of VMMC into key health facility activities through the designation of the facility as a department yielded positive results in regard to resource supply, both human and material [14] .

The outputs of the program reveal a marked variation between the 2 districts. Whereas Seke performed exceptionally well, scoring 99.5% circumcisions during the year 2016, Goromonzi posted 48.5% success from their set target. The targets were set considering the population denominator and not in consultation with the implementers. The program aims to circumcise 1.3 million HIV-negative males in order to realise 85% reduction in HIV transmission. The criterion to allocate provinces and districts population specific targets was applied in order to realise the national goal.

Considering program performance from inception in 2009 to date, and its mission to circumcise 1.3 million HIV-negative males between 10 to 29 years of age, Seke is on course at 83%. A different situation is obtaining with Goromonzi which has achieved only 15% of their set target. If this situation is not mitigated, Goromonzi is set, not only to miss their district target but to also impact negatively on the provincial as well as the national performance. This revelation, therefore, points to an urgent need to address the situation in Goromonzi to enhance performance.

The most common reason cited by participants from Goromonzi for the underperformance of the VMMC program was a late disbursement of payments. On analysis, this reason is invalidated by the fact that disbursements were simultaneously done for Seke district as well. The funds are from the same partner at similar rates.